Improving Intraocular Lens Calculation in Fuchs Endothelial Dystrophy According To Preoperative Degree of Corneal Edema

Fuchs endothelial dystrophy (FED) is a bilateral, asymmetric, progressive corneal endothelium disorder that causes corneal edema. Resolution of corneal edema is only possible by corneal transplantation. Cataract surgery is a common surgery that replaces the natural lens of the eye by an articial intraocular lens (IOL). The IOL-power calculation depends mainly on the anterior corneal keratometry and the axial length. In patients with FED, anterior keratometry may be affected by corneal edema and calculations may be less accurate. Therefore, the aim of this study is to establish the theorical postoperative refractive error due to corneal edema resolution after Descemet stripping endothelial keratoplasty combined with cataract surgery and IOL implantation. For this, anterior keratometry was measure preoperatively with edematous cornea and postoperatively after corneal edema resolution. Both keratometries were compared and used to calculate the respective theorical IOL-powers. The difference between target IOLs was used to establish the theorical refractive error due to corneal edema resolution. The results showed that corneal edema resolution induces a change in anterior keratometry, which affects IOL-power calculations and causes a hyperopic shift. The patients with moderate-to-severe preoperative corneal edema had higher theorical refractive error so their target selection should be adjusted for additional − 0.50D.


Introduction
Fuchs endothelial dystrophy (FED) is a bilateral, asymmetric, progressive corneal endothelium disorder. This autosomal dominant pathology can cause blurry vision secondary to corneal edema, so corneal transplantation is necessary. 1 Intraocular lens (IOL) calculation is a problem when endothelial keratoplasty is combined with cataract surgery in patients with FED. All corneal changes induced by corneal edema can alter IOL calculation and cause higher residual refractive errors. [2][3][4][5][6] Descemet Stripping Automated Endothelial Keratoplasty (DSAEK) has become the most common endothelial keratoplasty for treating endothelial dysfunction in the last decade because it provides, among other advantages, early visual recovery. 7 Some reports previously advocate that this surgery induces a postoperative mean hyperopic shift of + 1.10D (from + 0.70D to + 1.50D), 7,8 attributed to the concave shape of the donor corneal endothelial lenticule. [9][10][11] However after Descemet Membrane Endothelial Keratoplasty (DMEK), a minor hyperopic shift has been described of + 0.31D (from + 0.03D to + 1.20D). 7 In DMEK cases, the hyperopic shift cannot be explained by donor lenticule because this is thinner. Hence the concave shape of the donor lenticule is eliminated to prevent postoperative refractive change. It has been suggested that the hyperopic shift could be caused by structural corneal changes associated with corneal edema ( attening of the central posterior keratometry, oblate posterior cornea shape and steepening anterior corneal surface). 2,3,5,6,12,13 Surgeons frequently increase the IOL power in endothelial keratoplasty combined with phacoemulsi cation and IOL implantation (triple-DSAEK or triple-DMEK) to minimize this postoperative mean refractive error (RE). However, a high standard deviation appears in the outcomes that can be explained by differences in the preoperative degree of corneal edema in each patient. The purpose of this study is to quantify the IOL power calculation error (EIOL) due to corneal edema resolution after triple-DSAEK and the RE related to this effect. To the best of our knowledge, our study is the rst to estimate the RE according to the degree of preoperative corneal edema. These outcomes will allow more accurate IOL calculations in combined procedures of endothelial keratoplasty.

Results
Thirty-four eyes of 34 patients, of whom 23 (68%) were female and 11 (32%) male, were included. The average patient age was 65.44±9.45 years (range: 49-82 years). All the patients were Caucasian, diagnosed with FED and underwent triple-DSAEK. Figure 1 show the average preoperative keratometry (K1 or at corneal keratometry, K2 or steep corneal keratometry and Km or mean corneal keratometry) with edematous cornea (K preDSAEK ) compared to the average postoperative keratometry after corneal edema resolution (K postDSAEK ).

Intraocular Lens Power Calculations
The power of the IOLs was calculate using true keratometries in a theorical eye with 24mm axial length.
However, RE was statistically independent of the preoperative keratometry with corneal edema (RE & K preDSAEK , P > 0.05 in all cases)

Results According to the Preoperative Degree of Corneal Edema
The sample was divided into two groups according to the degree of preoperative corneal edema. Seventeen eyes (50%) were classi ed as "mild edema", and 17 eyes (50%) as "moderate-to severe" edema. The patients from both groups had similar gender, age and K preDSAEK characteristics (P > 0.05 in all the comparisons). Table 1 shows the keratometry error after corneal edema resolution (EK = K preDSAEK -K postDSAEK ) according to the preoperative degree of corneal edema.
The error observed in the IOL calculation (EIOL) varied depending on the degree of preoperative corneal edema. Thus, IOL power was more underestimated in the patients with moderate-to-severe preoperative corneal edema than in those with mild edema (-1.09±0.77D and -0.29±0.56D, respectively; P = 0.002, 95% CI 0.32-1.26 D).
As a result, theorical RE varied according to the degree of preoperative corneal edema ( Figure 2). The average hyperopic RE was +0.52D greater in the patients with moderate-to-severe preoperative corneal edema than in those with mild preoperative corneal edema (+0.71±0.52D and +0.19±0.37D, respectively; P = 0.003, 95% CI 0.21-0.84). Table 2 summarizes the preoperative and postoperative data of all the patients, the degree of corneal edema in each case and the RE induced by corneal edema resolution. The preoperative data include the K-readings (K1, K2 and Km) with corneal edema (K preDSAEK ), the theorical IOL power calculated with corneal edema (IOL preDSAEK ) and the selected target of IOL preDSAEK .
The postoperative data include K-readings after corneal edema resolution (K postDSAEK ), the theorical IOL power calculated after corneal edema resolution (IOL postDSAEK ) and the selected target of IOL postDSAEK . The selected target on both IOLs was similar (P = 0.737).

Discussion
This study observed that triple-DSAEK surgery induced the postoperative attening of anterior keratometry due to corneal edema resolution after surgery. These changes in postoperative anterior keratometry alter IOL calculations and cause a hyperopic shift.
A hyperopic shift after endothelial keratoplasty is well-known and is commonly explained by changes in the posterior corneal surface. After DSAEK, the concave shape of the donor lenticule induced a negative RE by diminishing the total refractive corneal power. 8,9,11,15 However, only 2/3 of the RE can be attributed to the shape and size of the donor lenticule. 11 After DMEK, the thin donor tissue had no refractive power, so it was unable to justify the hyperopic shift. Even so, it has been demonstrated that postoperative changes in curvature and asphericity of the posterior cornea surface occur after DMEK and reduce the total refractive cornea power. 2,3,12,16,17 However, the changes found in the posterior corneal surface appeared insu cient to justify the reported total hyperopic shift. Recently, some authors have observed that patients with a more marked hyperopic shift after triple-DMEK showed similar preoperative corneal features: corneal posterior surface keratometry was atter, posterior asphericity was positive and pachymetry was thicker than in normal corneas. 5,6,17 Furthermore, all authors agree that the changes observed before and after transplant are secondary to changes in the degree of corneal hydration. Thus corneal changes due to edema in FED may determine IOL calculation accuracy in procedures that combine endothelial transplant, phacoemulsi cation and IOLimplant (triple procedures). 3,16 The main IOL calculation devices based on partial coherence interferometry, such as IOL-Master, cannot measure the posterior cornea surface. These devices base their results on third-generation formulas in which the main variables are anterior cornea keratometry and AXL. So if anterior cornea keratometry is not altered by corneal hydration as some authors suggest, 3 In our study, the same device was used to measure the preoperative and postoperative keratometries (K preDSAEK and K postDSAEK , respectively). IOL Master 500 showed statistically signi cant keratometric attening of approximately 0.60D after edema resolution, which is similar to previous data published by other authors. [2][3][4]19 Therefore, it could be assumed that the corneal edema present in patients with FED overestimates keratometry.
In order to verify the effect of corneal edema on IOL-power selection, we performed IOL calculation twice: once using the initial preoperative keratometry (with corneal edema) and a second with the postoperative keratometry (after edema resolution). We performed these two IOL-power calculations in the same theorical model eye to eliminate other causes of error in IOL calculations (such as effective lens position, AXL or formula). Thus, we can verify the direct relation between keratometric change (variable EK) and the error in IOL-power selection (EIOL).
Our results showed that, after edema resolution, IOL-power (IOL postDSAEK ) was approximately 1D higher than that calculated with corneal edema (IOL preDSAEK ). This difference in IOL calculation occurred because the overestimated preoperative keratometry (K preDSAEK ) conditioned insu cient IOL-power (IOL preDSAEK ). Finally, the underestimated IOL (IOL preDSAEK ) generated a mean theorical hyperopic shift of + 0.45D after corneal edema resolution (with K postDSAEK ). Curiously, the hyperopic error obtained herein was similar to previously published data in a report by the American Academy of Ophthalmology after DMEK (+ 0.31D). 7 It should be considered that this hyperopic shift is calculated for a theoretical eye with an AXL of 24 mm. Thus, shorter eyes with higher IOL-power will show greater hyperopic errors, while we observed the opposite in longer eyes.
This study is the rst one to establish a direct relation between hyperopic shift and the degree of preoperative corneal edema in patients with FED. To date, different authors have theorized that the corneal changes responsible for hyperopic shifts are secondary to variations in the corneal hydration level. However, none has measured or quanti ed the preoperative corneal edema level. To date, authors have only related changes after endothelial keratoplasty to pachymetric changes 5,20 , variations in corneal densitometry 13 or preoperative positive asphericity on the posterior corneal surface 6 . In the present study, we divided our sample in two groups according to the degree of preoperative corneal edema observed by slit lamp (both groups were similar in number, age and preoperative keratometry terms). Data demonstrate that different degrees of preoperative corneal edema induce statistically distinct results.
Speci cally, the patients undergoing triple-DSAEK with moderate-to-severe preoperative edema showed more marked keratometry attening after edema resolution than those with mild preoperative edema. Consequently, the hyperopic shift in the patients with more corneal edema was + 0.52D greater (+ 0.71 ± 0.52D vs + 0.19 ± 0.37D, respectively). Therefore, we suggest that the RE related to corneal edema should be considered in IOL-power calculations. Increasing the negative target of IOL according to the preoperative degree of edema could diminish postoperative refractive surprises after triple procedures.
Most surgeons choose a negative IOL target to compensate the hyperopic shift; from − 0.50D to -1.00D in triple-DMEK 3,5,6,21−23 and from − 0.8D to -1.25D in triple-DSAEK. [24][25][26] According to our results, we propose selecting the − 0.19D IOL target in the patients with mild edema and one of -0.71D IOL in those patients with moderate-to-severe edema in triple-DMEK. In triple-DSAEK, an additional myopic target from − 0.62D to -0.97D, secondary to the effect of the donor lenticule, must be added. 10,11 So in triple-DSAEK, we propose selecting an IOL target from − 0.81D to -1.16D and from − 1.33D to -1.68D in the patients with mild edema and with moderate-to-severe edema, respectively.
When the degree of preoperative corneal edema is not evaluated, choosing a -0.45D myopic target could compensate an average error induced by corneal edema resolution, but with wider variability in the results. This average adjustment in IOL calculation agrees with that suggested by other authors. 6,20 .  20 These authors propose adding − 0.50D of myopic adjustment of the target in those eyes with a preoperative corneal central thickness of 640 µm or more. With the proposed adjustment, their mean postoperative refractive surprise dropped from + 1.20 ± 0.92 to + 0.70 ± 0.92 D. However, corneal central thickness is extremely variable between patients and may not be universally valid for de ning the IOL-power target. 17,28,29 So once again, corneal transparency evaluation by slit lamp could be more reliable for establishing the degree of corneal edema and for setting the IOL target.
Some limitations in the current study should be taken into account. The subjective classi cation of the preoperative degree of corneal edema may reduce the repeatability of our results. However, we currently have no effective objective systems for grading corneal edema. To reduce the variability of our classi cation, only two well-differentiated groups were recorded: a mild corneal edema group including clear corneas and moderate-to-severe corneal edema included unclear corneas that do not allow details of anterior chamber to be seen. Besides, corneal irregularities secondary to FED severity could make keratometry acquisition di cult. 29 In our study, only repeatable keratometries were recorded. IOL-Master 500 performs keratometries in 2.3 mm of the central cornea. 30 Thus, peripheral corneal irregularities secondary to severe edema should not affect (or less affect) the recorded keratometry and our results.
In summary, corneal edema resolution after a triple-DSAEK procedure induces an anterior keratometry attening of -0.64 D.
This corneal attening induces an error in IOL calculations, which is the cause of an average hyperopic shift of + 0.45D.
Moderate-to-severe preoperative corneal edema presented a greater hyperopic shift. Thus, we advocate adding − 0.50D more myopic IOL target in the patients with moderate-to-severe preoperative corneal edema than in those with preoperative mild edema.

Surgical Technique
All surgical procedures were performed by a single surgeon using peribulbar or topical anesthesia. Cataract surgery was carried out with conventional phacoemulsi cation and IOL implantation in the capsular bag. DSAEK was performed following the cataract procedure. The DSAEK surgery procedure has been previously described by Ortiz-Gomariz et al. 1414 Recorded Data Patient demographics, the degree of preoperative corneal edema, preoperative keratometry (K preDSAEK ) and postoperative keratometry (K postDSAEK ) were recorded. The degree of preoperative corneal edema was subjectively evaluated by a corneal surgeon based on slit lamp ndings. Mild edema was de ned as minimal microcystic corneal edema in patients in early FED stages. In this stage, corneal clearance must allow the perfect visualization of iris and lens details. Moderate-to-severe edema was de ned as evident stromal corneal edema in the patients with advanced FED. This advanced degree of corneal edema does not allow the perfect visualization of iris or lens details. Postoperative keratometry was recorded at least 6 months after surgery after complete edema resolution (the cornea should be clear, and the BCVA should be 0.1 LogMAR or better).
The preoperative and postoperative keratometry (K) data were obtained by IOL-Master 500 (Carl Zeiss, Germany). Each measure was repeated 3 times and the average was recorded. Each K-reading included K1 ( at corneal keratometry), K2 (steep corneal keratometry) and Km (mean corneal keratometry).
Preoperative and postoperative K-readings were compared. Keratometry error (EK) was calculated by the difference between preoperative keratometry and postoperative keratometry (EK = K preDSAEK -K postDSAEK ). Positive values suggest an overestimation of the preoperative K-reading due to corneal edema, while negative values imply underestimated preoperative K-readings.

Intraocular Lens Power Calculation
Real K-readings were used for calculated two intraocular lens (IOL) in a theorical eye model with an axial length (AXL) of 24 mm. Thus we obtained IOL preDSAEK using preoperative keratometry (K preDSAEK ), and we also calculated IOL

Statistical analysis
Data distribution normality was veri ed by the Kolmogorov-Smirnov test. The data before and after surgery were compared by a paired t test. RE was calculated and correlated with the other parameters by a Rho-Spearman correlation test. Finally, the sample was divided into two groups according to the degree of preoperative corneal edema. The statistical differences between the two groups, de ned by the preoperative degree of corneal edema ("mild edema" and "moderate-to-severe edema"), were determined by an unpaired t test (parametric variables: age, keratometry, IOL power) and a U-Mann-Whitney test (nonparametric variables: RE). A P-value ≤ 0.05 was considered statistically signi cant.
Declarations ACKNOWLEDGEMENT This study has been supported in part by the Red Temática de Investigación Cooperativa en Salud (RETICS), reference number RD16/0008/0012, Funded by Instituto de Salud Carlos III and co-funded by European Regional Development Fund (ERDF), "A way to make Europe".   All patient data are shown: Degree of corneal edema; preoperative data that include k-readings with corneal edema (K preDSAEK ); IOL-power calculated with K preDSAEK (IOL preDSAEK ) and predicted spherical equivalent obtained with IOL preDSAEK ; postoperative data that include postoperative K-readings after corneal edema resolution (K postDSAEK ), IOLpower calculated with K postDSAEK (IOL postDSAEK ) and predicted spherical equivalent obtained with IOL postDSAEK ; and theorical refractive error induced by IOL preDSAEK secondary to keratometric change after postoperative corneal edema resolution.  Figure 1 Preoperative K-readings with corneal edema (KpreDSAEK) versus postoperative K-readings (KpostDSAEK) without corneal edema. * Statistically signi cant data (P ≤ 0.001) by a T-Student-Paired-Sampled test Figure 2